/*---------------------------------------------------
 * file:    docword2dw.c
 * purpose: 
 * usage:   
 * inputs:  docword.txt P
 * outputs: dw.p0.bin ... dwz.p7.bin
 * version: 1.0
 * author:  newman@uci.edu
 * date:    8/10/07
 * NOTE:    binary matrices are all (c-style) zero offset
 *-------------------------------------------------*/

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <math.h>
#include <assert.h>
#include "topiclib.h"

/*==========================================
 * main
 *========================================== */
int main(int argc, char* argv[])
{
  int P, D, W, N, NNZ;
  int i, j, k, chk, n2=2;
  int *dd, *ww, *cc, *d, *w;
  int Dp, p, dstart, dend, nstart, nend, Np;
  int *dSplit, *wSplit;
  FILE *fin, *fout;
  char fname[32];

  if (argc == 1) {
    fprintf(stderr, "usage: %s P\n", argv[0]);
    exit(-1);
  }
  P = atoi(argv[1]);  assert(P>0);

  /*==========================================
   * read docword.txt
   *========================================== */
  fin = fopen("docword.txt","r");        assert(fin);
  fscanf(fin,"%d", &D);                  assert(D>0);
  fscanf(fin,"%d", &W);                  assert(W>0);
  fscanf(fin,"%d", &NNZ);                assert(NNZ>0);
  dd = ivec(NNZ);                        assert(dd);
  ww = ivec(NNZ);                        assert(ww);
  cc = ivec(NNZ);                        assert(cc);
  for (j = 0; j < NNZ; j++)
    fscanf(fin,"%d%d%d", dd+j, ww+j, cc+j);
  fclose(fin);
  printf("... read docword.txt\n");
  printf("       D   = %d\n", D);
  printf("       W   = %d\n", W);
  printf("       NNZ = %d\n", NNZ);
  
  /*==========================================
   * allocate memory d and w
   *========================================== */
  N = isum(NNZ,cc);
  d = ivec(N);                           assert(d);
  w = ivec(N);                           assert(w);
  printf("       N   = %d\n", N);

  /*==========================================
   * assign d and w
   *========================================== */
  i = 0;
  for (j = 0; j < NNZ; j++) {
    for (k = 0; k < cc[j]; k++) {
      d[i] = dd[j]-1;
      w[i] = ww[j]-1;
      i++;
    }
  }
  assert(i==N);
  free(dd);
  free(ww);
  free(cc);
  printf("... assigned d and w\n");

  /*==========================================
   * DISTRIBUTE over P datafiles
   *========================================== */
  Dp = D/P;
  nstart = 0;
  for (p = 0; p < P; p++) {
    
    // Compute offsets
    dstart =  p   *Dp;
    dend   = (p+1)*Dp;  if (p==(P-1)) dend=D;
    nend   = nstart;
    while( (d[nend]<dend) && (nend<N) ) nend++;
    printf("... p=%d dstart=%d dend=%d nstart=%d nend=%d Np=%d\n", p, dstart, dend, nstart, nend, nend-nstart);
    
    // Allocate
    Np     = nend - nstart;
    dSplit = ivec(Np);   assert(dSplit);
    wSplit = ivec(Np);   assert(wSplit);
    
    for (i = nstart; i < nend; i++) {
      dSplit[i-nstart] = d[i] - dstart;
      wSplit[i-nstart] = w[i];
    }
    
    // Write
    sprintf(fname,"dw.p%d.bin",p);
    fout = fopen(fname,"w");                   assert(fout);
    chk = fwrite(&n2, sizeof(int),1,fout);     assert(chk==1);
    chk = fwrite(&Np, sizeof(int),1,fout);     assert(chk==1);
    chk = fwrite(dSplit,sizeof(int),Np,fout);  assert(chk==Np);
    chk = fwrite(wSplit,sizeof(int),Np,fout);  assert(chk==Np);
    fclose(fout);
    printf("       wrote %s\n", fname);
    
    free(dSplit);
    free(wSplit);

    nstart = nend;

  }

  return 0;
}